Genomic prediction by single-step genomic BLUP using cow reference population in Holstein crossbred cattle in India

TY - GEN

T1 - Genomic prediction by single-step genomic BLUP using cow reference population in Holstein crossbred cattle in India

AU - Nayee, Nilesh Kumar

AU - Su, Guosheng

AU - Gajjar, Swapnil

AU - Sahana, Goutam

AU - Saha, Sujit

AU - Trivedi, Kamlesh R.

AU - Guldbrandtsen, Bernt

AU - Lund, Mogens Sandø

N1 - Conference code: 11

PY - 2018

Y1 - 2018

N2 - Advantages of genomic selection in breeds with limited numbers of progeny tested bulls have been demonstrated by adding genotypes of females to the reference population (Thomasen et al., 2014). The current study was conducted to explore the feasibility of implementing genomic selection in a Holstein Friesian crossbred population with cows kept under small holder conditions using test day records and single step genomic BLUP (ssGBLUP). Milk yield records from 10,797 daughters sired by 258 bulls were used Of these 2194 daughters and 109 sires were genotyped with customized genotyping chip on the Illumina platform. A conventional test day random regression model with Legendre Polynomials of 3rd order for both fixed and random regression was compared with an ssGBLUP model with same structure and effects as the conventional model. The conventional test day model and the ssGBLUP model were assessed using a 5-fold cross validation in two scenarios. In scenario 1, half-sibs in the whole data were randomly divided into 5 subsets. In scenario 2, individuals in the whole data were randomly divided into 5 subsets. The predictive ability of the two models was assessed for accuracy of EBV. Accuracy was calculated as the correlation between 305 day EBV and corrected 305 day phenotypic values (Yc). Validation results showed that genomic prediction using ssGBLUP led to higher accuracy than the conventional test-day model. In validation Scenario 2 where validation animals had a relatively close relationship with training animals, there was a marginal increase of correlations in genotyped animals from 0.397 to 0.405. The difference between genomic and conventional prediction was more marked in validation scenario 1 where validation animals had more distant relationship with training animals. Correlations increased from 0.150 to 0.194 for all animals and from 0.247 to 0.276 for genotyped animals. In addition, correlation between average daughter Yc and predicted breeding values of sires increased from 0.126 when using conventional BLUP to 0.202 when using ssGBLUP. These results indicate that genomic selection is feasible in Indian HF Crossbred cattle even though only recent female animals have been genotyped. It is expected that the prediction accuracy will increase further as more animals are genotyped. Keywords: test day random regression model, ssGBLUP, genomic selection, Holstein Crossbred cows, small holder.

AB - Advantages of genomic selection in breeds with limited numbers of progeny tested bulls have been demonstrated by adding genotypes of females to the reference population (Thomasen et al., 2014). The current study was conducted to explore the feasibility of implementing genomic selection in a Holstein Friesian crossbred population with cows kept under small holder conditions using test day records and single step genomic BLUP (ssGBLUP). Milk yield records from 10,797 daughters sired by 258 bulls were used Of these 2194 daughters and 109 sires were genotyped with customized genotyping chip on the Illumina platform. A conventional test day random regression model with Legendre Polynomials of 3rd order for both fixed and random regression was compared with an ssGBLUP model with same structure and effects as the conventional model. The conventional test day model and the ssGBLUP model were assessed using a 5-fold cross validation in two scenarios. In scenario 1, half-sibs in the whole data were randomly divided into 5 subsets. In scenario 2, individuals in the whole data were randomly divided into 5 subsets. The predictive ability of the two models was assessed for accuracy of EBV. Accuracy was calculated as the correlation between 305 day EBV and corrected 305 day phenotypic values (Yc). Validation results showed that genomic prediction using ssGBLUP led to higher accuracy than the conventional test-day model. In validation Scenario 2 where validation animals had a relatively close relationship with training animals, there was a marginal increase of correlations in genotyped animals from 0.397 to 0.405. The difference between genomic and conventional prediction was more marked in validation scenario 1 where validation animals had more distant relationship with training animals. Correlations increased from 0.150 to 0.194 for all animals and from 0.247 to 0.276 for genotyped animals. In addition, correlation between average daughter Yc and predicted breeding values of sires increased from 0.126 when using conventional BLUP to 0.202 when using ssGBLUP. These results indicate that genomic selection is feasible in Indian HF Crossbred cattle even though only recent female animals have been genotyped. It is expected that the prediction accuracy will increase further as more animals are genotyped. Keywords: test day random regression model, ssGBLUP, genomic selection, Holstein Crossbred cows, small holder.

M3 - Article in proceedings

VL - 11

BT - Proceedings of the World Congress on Genetics Applied to Livestock Production, 2018

T2 - The 11th World Congress on Genetics Applied to Livestock Production

Y2 - 11 February 2018 through 16 February 2018

ER -